PROJECT SUMMARY A high-velocity eye movement or saccade is typically the first motor action we make to orient to an object of interest. While the neural mechanisms of saccade generation to stationary targets have been thoroughly investigated, very little is known about the neural control of interceptive saccades that acquire moving targets. Current dogma based on studies of saccades to stationary targets states that the visual and motor bursts in the superior colliculus (SC), a major hub in the oculomotor neuraxis, are represented as Gaussians; that the population activity is centered at the site encoding the target location and, equivalently, desired saccade vector; that its width remains invariant across different target locations and saccade vectors; and that these spatial features emerge from a balance of excitation and inhibition mediated through intrinsic, intra-laminar connectivity. Fundamentally non-overlapping mechanisms must be involved when the target is moving, because accurate interception can only occur if target velocity information is incorporated in the saccade command. We reason that as a moving target?s image streaks across the retina, activity sweeps across the SC too. We hypothesize that the population activity, which starts as a Gaussian to represent the initial visual response, becomes skewed as it sweeps across the SC; that the extent to which SC population activity is modified depends on the intra-laminar connectivity weights, the logarithmic map of visual space in SC, and target speed; that the altered spatial distribution persists during the peri-movement burst; and that an appropriate computational algorithm must be able to decode the saccade goal from the skewed population response. We propose to test these hypotheses using a combination of experimental and computational approaches. Specific Aim 1 will employ an innovative method for simultaneously recording neural activity of many SC neurons within a functional layer in nonhuman primates performing oculomotor tasks and compare the spatiotemporal properties of population activity during saccades to stationary and moving targets (different speeds and directions). Specific Aim 2 will construct a computational model that simulates population activity in SC and associated saccades to stationary and moving targets. We will employ a distributed architecture for the superficial and deeper layers of the SC and a lumped block-diagram circuit for the brainstem burst generator elements, like that done by Arai and colleagues (Neural Networks, 7:1115-1135, 1994). Collectively, these projects will provide an in-depth insight into the mechanisms for generation of interceptive saccades and enable a comparison with mechanisms of saccades to stationary targets.